Individual-Level Targeting

Transcription

1 Individual-Level Targeting Individual-level targeting. Review choice modeling framework. Using choice models for individual-level targeting (predictive modeling). Demonstration of software for BBBC case. Arvind Rangaswamy MKTG 521- Spring

2 Three Approaches to Targeting Macro-level targeting (e.g., segment selection as in Pacific Brands) GE-McKinsey Portfolio Matrix Micro-level targeting using Classification models Discriminant function CART Classification and Regression Trees Micro-level targeting using RFM, Regression, Choice Models, Neural networks, Data mining, etc. MKTG 521- Spring

3 Models for Targeting Individual Customers Step 1 Step 2 Step 3 Step 4 Step 5 Create database of customer responses (choices) based either on test mailing to a sample of prospects/customers, or historical data of past customer purchases. Use models such as regression, RFM, and Logit to assess the impact of independent variables (drivers) of customer response. Score each customer/prospect based on the drivers identified in Step 2 - the higher the score, the more likely is the predicted response. Classify customers into deciles (or smaller groupings) based on their scores. Based on profitability analyses, determine the top deciles to which a marketing action (e.g., mailing of brochure) will be targeted. MKTG 521- Spring

4 Step 1 Database for BookBinders Book Club Case Predict response to a mailing for the book, Art History of Florence, based on the following variables accumulated in the database and the responses to a test mailing: Gender Amount purchased Months since first purchase Months since last purchase Frequency of purchase Past purchases of art books Past purchases of children s books Past purchases of cook books Past purchases of DIY books Past purchases of youth books MKTG 521- Spring

5 Attributes in ABB s Choice-Segmentation Model Invoice price Energy losses Overall product quality Availability of spare parts Clarity of bid document Knowledgeable salespeople Maintenance requirement Ease of installation Warranty MKTG 521- Spring

6 Step 2 Drivers of the RFM Model R Recency Time/purchase occasions since the last purchase F Frequency Number of purchase occasions since first purchase M Monetary Value Amount spent since the first purchase Total RFM Score: R Score + F score + M Score MKTG 521- Spring

7 Step 2 Example RFM Model Scoring Criteria R Months from last purchase 13- max Max Score = 50 F Frequency > Max Score = 30 M Amount purchased Max Score = 20 > Implement RFM model using Nested If statements in Excel MKTG 521- Spring

8 Customer Needs and Customer Value Measurement Customer Needs and Buying Process Behaviors Present State Desired State Functional and Economic Needs Motivation Perceived and Psychological Needs Ignore Postpone Engage in Purchase Process Search for options Evaluate options Choose product Purchase product Use product Customer Value Measurement Approaches Objective Measures of Value Perceptual Measures of Value Behavioral Measures of Value

9 Choice Models vs Needs Surveys With standard survey methods... preference/ importance choice weights perceptions predict observe/ask observe/ask But with choice models... importance choice weights perceptions observe infer observe/ask MKTG 521- Spring

10 Using Choice Models to Determine the Value of a Customer 1. Observe choice: Buy/not buy -- direct marketers Brand bought -- packaged goods 2. Capture related characteristics data: demographics attitudes/perceptions market conditions (price, promotion, etc.) pattern of previous choices 3. Link 1 to 2 via choice model the model predicts customers probabilities of purchase and also reveals importance weights of characteristics. MKTG 521- Spring

11 Step 2 Logit Model of Response to Direct Mail Probability of responding to direct mail solicitation function of (past response behavior, = marketing effort, characteristics of customers) MKTG 521- Spring

12 Step 2 The Multinomial Logit Choice Model (Marketing Engineering Software) The primary objective of the model is to predict the probabilities that an individual will choose each of several choice alternatives. The model has the following properties: The probabilities lie between 0 and 1, and sum to 1. The model is consistent with the proposition that customers pick the choice alternative that offers them the highest utility (value) on a purchase occasion, but the utility has a random component that varies from one purchase occasion to the next. The model has the proportional draw property -- each choice alternative draws from other choice alternatives in proportion to their utility. MKTG 521- Spring

13 Step 2 Conceptual Background for Choice Model U ij = V ij + e ij Value of Buying Art History of Florence Value for = + Gender Value for previous purchase of art books Value of unknown factors where: U ij = Utility (or hidden value) that customer i has for choice alternative j. In Bookbinder s case, choice alternatives are Buy and Don t Buy. We can set j = 1 for Buy and j = 0 for Don t Buy. We do not observe U ij. For identification, we will also set U ij = 0 for Don t Buy V ij = Deterministic (or observable) component of utility that is a function of several attributes associated with choice alternatives. e ij = An error term that reflects the non-deterministic component of utility. MKTG 521- Spring

14 Step 2 Specification of the Deterministic Component of Utility V ij = K k=1 w k X ij k where: i = an index to represent customers. k k X ij = an index to represent an independent variable. = The values of the K independent variables for customer i for choice alternative j. w k = estimated coefficient to represent the impact of X k ij on the value (utility) realized by customer i. MKTG 521- Spring

15 The Special Case of Binary (Two Alternatives) Logit Model Step 2 If customers maximize the value associated with their choices and the non-deterministic component error (e i ) has a specific form called double exponential, then: where: p i1 = ev i1 e 0 + e V i1 = ev i1 1 + e V i1 p i1 = predicted probability that customer i will respond (j = 1) to the direct mail. V i1 = estimated value of the offering to customer i (i.e., based on estimates of w k ) obtained from maximum likelihood estimation. It varies from - (minus infinity) to. The value to the customer of not having the offering (j = 0) is set to 0 (base value). MKTG 521- Spring

16 More General Case (Multinomial Logit Model) p ij = ev ij J j=1 e V ij If there are three choice alternatives, we will get three probability predictions: p i1 = ev i1 3 j=1 e V ij ; p i2 = ev i2 3 j=1 e V ij ; p i3 = ev i3 3 j=1 e V ij MKTG 521- Spring

17 Step 2 An Important Property of the Logit Model The marginal impact is highest when the customer is sitting on the fence, i.e., when p i = 0.5. Marginal impact of variable j (e.g., price) on p i High Low Probability of Individual i Choosing the product p i Question: Is this a good property to have? MKTG 521- Spring

18 Step 3 Computing Scores Based on Regression Run regression model to predict probability of purchase: Choice (0 or 1) = Function of variables (e.g., Gender) + error i.e., the estimated regression equation is given by: P ij = w 0 + k w k X ijk w's are estimated regression coefficients. P ij is the probability that individual i will choose alternative j. Note that predicted choice probabilities from the regression model need not necessarily lie between 0 and 1, although most of the probabilities will fall in that range. MKTG 521- Spring

19 Step 3 Compute Prediction Scores from Different Models RFM Score: Use computed score as an index of the probability of purchase (i.e., higher the RFM score, the greater the probability of purchase). Regression: Score ( for customer i) wˆ 0 wˆ k X ijk k Logit: Customer i' s score (probability) e wˆ 0 1 e wˆ wˆ 0 k X wˆ k ijk X ijk w's are weights estimated by the Regression or Logit models. RFM and Regression models can be implemented in Excel. Also, all three scoring procedures for probability of purchase can be implemented in Excel. MKTG 521- Spring

20 Step 3 Score Customers for their Potential Profitability (Example) A B C D Score Average Customer (Purchase Purchase Expected $ Customer Probability) Volume Margin = A B C 1 30% $ % $ % $ % $ % $ % $ % $ % $ % $ % $ Average expected purchase per customer = $3.72 MKTG 521- Spring

21 Step 4 Decile Classification Example Decile Customer(s) $ If the marketing cost to reach a customer is $3, at what decile will you stop your targeting effort? How is this targeting plan different from one based on average purchases of customers ($3.72)? MKTG 521- Spring

22 Step 4 Decile Classification Standard Assessment Method Apply the results of approach and calculate the score of each individual (calibration versus test sample). Decile1 Customer 1 Score 1.00 Customer 2 Score Customer 230 Score 0.92 Order the customers based on score from the highest to the lowest. Divide into deciles..... Calculate/graph hit rate and profit. Decile10 Customer 2300 Score 0.00 MKTG 521- Spring

23 Step 5 Determine Targeting Plan (Example shows potential profitability of top 6 deciles) Model RFM Number of hits (favorable responses at 60 th percentile of ordered scores) Expected response rate by mailing the top 60% of customers in the ordered list % of likely responders recovered at 60 th percentile Regression MNL Compute profit/roi for the models based on the number of mailings recommended by each model and compare that to mailing to the entire list (equivalently to a randomly selected list of the same size). MKTG 521- Spring

24 Step 5 Develop Lift Charts and Choose Model for Implementation Cumulative Hit Rate (%) 100% 80% 60% 40% 20% 0% Hit Rate Random Hit Profit Random Prof Decile 55,000 50,000 45,000 40,000 35,000 30,000 25,000 20,000 15,000 10,000 5,000 0 Cumulative Profit ($) MKTG 521- Spring

25 Database Approach to Implementing Individual Targeting Appended Company Data Sales, customer support, shipments, payments, etc. Model-based Scores Third-Party Data (e.g., Credit Scores) Operational Database Customer Database Transactions General Ledger Offers and Responses MKTG 521- Spring